Environment International 131 (2019) 104988 Contents lists available at ScienceDirect Environment International journal homepage: www.elsevier.com/locate/envint Exposure to disinfection by-products in swimming pools and biomarkers of T genotoxicity and respiratory damage – The PISCINA2 Study Laia Font-Riberaa,b,c,d, Esther Marcoe, Joan O. Grimalte, Susana Pastorc,f, Ricard Marcosc,f, Lilianne Abramsson-Zetterbergg, Marie Pedersenh,i, Tamara Grummtj, Ralf Junekj, Esther Barreirob,k,l, Dick Heederikm, Jack Spithovenm, Rossana Critellin,o, Alessio Naccaratin, Christina Schmalzp, Christian Zwienerp, Jiaqi Liuq, Xiangru Zhangq, William Mitchr, Esther Gracia-Lavedana,b,c, Lourdes Arjonaa,b,c,d, Jeroen de Bonta,b,c, Lluïsa Tarèsa,b,c, Paolo Vineiss, Manolis Kogevinasa,b,c,d, Cristina M. Villanuevaa,b,c,d, ⁎ a ISGlobal, Barcelona, Spain b Universitat Pompeu Fabra (UPF), Barcelona, Spain c Consortium for Biomedical Research in Epidemiology and Public Health (CIBERESP), Carlos III Institute of Health, Madrid, Spain d IMIM (Hospital del Mar Medical Research Institute), Barcelona, Spain e Institute of Environmental Assessment and Water Research (IDÆA-CSIC), Barcelona, Spain f Universitat Autònoma de Barcelona (UAB), Barcelona, Spain g National Food Administration, Uppsala, Sweden h Department of Public Health, Section of Epidemiology, University of Copenhagen, Denmark i Danish Cancer Society Research Center, Diet, Genes and Environment, Copenhagen, Denmark j German Environment Agency, Bad Elster, Germany k Pulmonology Department-Lung Cancer & Muscle Research Group, IMIM (Hospital del Mar Medical Research Institute), Barcelona, Spain l CIBER de Enfermedades Respiratorias (CIBERES), Barcelona, Spain m Institute for Risk Assessment Sciences (IRAS), Utrecht, the Netherlands n Italian Institute for Genomic Medicine (IIGM), Torino, Italy o Department of Medical Science, University of Turin, Turin, Italy p Environmental Analytical Chemistry, Center for Applied Geoscience, University of Tübingen, Tübingen, Germany q Department of Civil and Environmental Engineering, Hong Kong University of Science and Technology, Hong Kong, China r Department of Civil and Environmental Engineering, Stanford University, Palo Alto, CA, USA s School of Public Health, Imperial College London, UK ARTICLE INFO ABSTRACT Handling Editor: Olga-Ioanna Kalantzi Background: Swimming in pools is a healthy activity that entails exposure to disinfection by-products (DBPs), Keywords: some of which are irritant and genotoxic. Biomarkers Objectives: We evaluated exposure to DBPs during swimming in a chlorinated pool and the association with CC16 short-term changes in genotoxicity and lung epithelium permeability biomarkers. Disinfection by-products Methods: Non-smoker adults (N = 116) swimming 40 min in an indoor pool were included. We measured a Exposure range of biomarkers before and at different times after swimming: trihalomethanes (THMs) in exhaled breath Genotoxicity (5 min), trichloroacetic acid (TCAA) in urine (30 min), micronuclei in lymphocytes (1 h), serum club cell protein Swimming pools (CC16) (1 h), urine mutagenicity (2 h) and micronuclei in reticulocytes (4 days in a subset, N = 19). Several DBPs in water and trichloramine in air were measured, and physical activity was extensively assessed. We estimated interactions with polymorphisms in genes related to DBP metabolism. Results: Median level of chloroform, brominated and total THMs in water was 37.3, 9.5 and 48.5, μg/L, respectively, and trichloramine in air was 472.6 μg/m3. Median exhaled chloroform, brominated and total THMs increased after swimming by 10.9, 2.6 and 13.4, μg/m3, respectively. Creatinine-adjusted urinary TCAA increased by 3.1 μmol/mol. Micronuclei in Abbreviations: BMI, body mass index; Br-THMs, brominated THMs; CYP2E1, cytochrome P450 2E1 gene; CC16, club cell protein; CC16, club cell protein gene; DBP, disinfection by-product; GSTT1, glutathione S-transferase theta 1 gene; GSTZ1, glutathione S-transferase zeta 1 gene; HAA, haloacetic acid; IQR, interquartile range; MHR, maximum heart rate; MN, micronuclei; MN-PBL, micronuclei in peripheral blood lymphocytes; MN-Ret, micronuclei in reticulocytes; SNP, single-nucleotide polymorphism; TCAA, trichloroacetic acid; THM, trihalomethane; TTHMs, total trihalomethanes ⁎ Corresponding author at: ISGlobal, Doctor Aiguader 88, Barcelona 08003, Spain. E-mail address: cristina.villanueva@isglobal.org (C.M. Villanueva). https://doi.org/10.1016/j.envint.2019.104988 Received 11 February 2019; Received in revised form 28 June 2019; Accepted 30 June 2019 Available online 16 July 2019 0160-4120/ © 2019 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/BY-NC-ND/4.0/). L. Font-Ribera, et al. Environment International 131 (2019) 104988 lymphocytes and reticulocytes, urine mutagenicity and serum CC16 levels remained unchanged after swimming. Spearman correlation coefficients showed no association between DBP exposure and micronuclei in lymphocytes, urine mutagenicity and CC16. Moderate associations were observed for micronuclei in reticulocytes and DBP exposure. Conclusions: The unchanged levels of the short-term effect biomarkers after swimming and null associations with personal estimates of exposure to DBPs suggest no measurable effect on genotoxicity in lymphocytes, urine mutagenicity and lung epithelium permeability at the observed exposure levels. The moderate associations with micronuclei in reticulocytes require cautious interpretation given the reduced sample size. 1. Introduction mutagenicity (Kogevinas et al., 2010) and serum CC16 (Font-Ribera et al., 2010). Several genes involved in DBP metabolism have been shown to Swimming is a healthy and highly practiced sport, but swimming in modify bladder cancer risk (Cantor et al., 2010) and were also evaluated in pools entails a considerable exposure to disinfection by-products (DBPs). the PISCINA study (Kogevinas et al., 2010). Hundreds of DBPs have been identified in swimming pool water The PISCINA study was the first to assess DBP genotoxicity among (Richardson et al., 2010), and some of them are genotoxic and carcino- volunteers swimming in a chlorinated pool and other similar studies genic in animals (Richardson et al., 2007). The major exposure route for have not been conducted. The assessment of short-term changes in re- volatile DBPs such as trihalomethanes (THMs) is inhalation (Erdinger spiratory biomarkers after swimming has been more studied, especially et al., 2004; Marco et al., 2015), whereas dermal absorption, accidental for lung epithelium permeability, with inconsistent results (Carbonnelle ingestion or inhalation of aerosol may be more relevant for other DBPs et al., 2002; Carbonnelle et al., 2008; Nordberg et al., 2012; Llana- (Cardador and Gallego, 2011). Exposure to THMs through drinking water Belloch et al., 2015). The role of physical activity is a key aspect in this and showering has been consistently associated with bladder cancer risk type of studies measuring biomarkers before and after swimming, (Villanueva et al., 2004). A study also suggested an increased risk of which has not been properly tackled in previous studies. bladder cancer with pool attendance (Villanueva et al., 2007). In addition, We aimed to assess short-term changes in biomarkers of genotoxicity indoor swimming pools accumulate volatile DBPs in the air, including and lung epithelium permeability in relation to DBP exposure in swimming irritant compounds such as trichloramine. Several studies have described a pools in a new study (PISCINA2) to replicate the previous PISCINA study in higher prevalence of respiratory disorders such as asthma among pool an independent and larger population. We used the effect biomarkers that workers and professional swimmers (Villanueva and Font-Ribera, 2012). were previously related to DBP exposure (serum CC16 for lung epithelium In order to provide evidence on the mechanisms of action of DBP ex- permeability and urine mutagenicity and MN-PBL for genotoxicity) and posure in swimming pools, we previously conducted an experimental study genotyped twelve polymorphisms in three genes involved in DBP metabo- (PISCINA study) in 50 healthy adults swimming during 40 min in an indoor lism. In addition, in this study we measured an extensive range of DBPs in chlorinated swimming pool (Kogevinas et al., 2010; Font-Ribera et al., water, air and biofluids, and used several indicators of physical activity. We 2010). Individual DBP exposure was measured as the THM concentration in finally also evaluated MN in reticulocytes (MN-Ret), which is a newer and exhaled breath after swimming. Several genotoxicity and respiratory bio- potentially more sensitive genotoxicity biomarker (Kissling et al., 2007). markers were assessed in biosamples collected before and at different times after swimming, including micronuclei (MN) and comet assay in peripheral 2. Methods blood lymphocytes, urine mutagenicity, MN in exfoliated urothelial cells in urine, exhaled nitric oxide, cytokines and 8-isoprostane in exhaled breath 2.1. Study design condensate, and surfactant protein D and Club cell protein (CC16) in serum. The individual exposure to bromoform, one of the most genotoxic THMs The PISCINA2 study included currently non-smoker and non-pro- (Richardson et al., 2007) measured as exhaled levels, was related to an fessional swimmers between 18 and 40 years old, recruited through increase in MN frequency in peripheral blood lymphocytes (MN-PBL), urine open advertisements at local research centres, universities, social Fig. 1. Study design and timing of sample collection. THMs: trihalomethanes; TCAA: trichloroacetic acid; MN: micronuclei; CC16: Club cell protein. DBPs: disinfection by-products. 2 L. Font-Ribera, et al. Environment International 131 (2019) 104988 media, and personal contacts. Participants were asked to avoid swim- (SD = 25,258). Only 19 participants had isolated reticulocytes for the ming pool attendance the week before, exercise three days before and before and after swimming sample and could be used for the analysis. up to three days after swimming (until the last sample collection), and showering in the morning of the study, and were provided with DBP- 2.2.3. Urine mutagenicity free bottled water during the study period. In total, 116 volunteers were Urine samples (30 mL) were collected before and 2 h after swim- recruited over 30 experimental days (June–December 2013), who were ming. Mutagenicity was evaluated with the Salmonella (Ames) muta- asked to swim during a fixed time of 40 min in a single indoor chlori- genicity plate-incorporation assay in strain YG1024 with S9 mix. The nated pool in Barcelona. Four participants were evaluated per day, with mutagenic potencies of samples, expressed as revertants per millilitre- start swimming times ranging from 10:00 a.m. to 11:42 a.m. equivalent (rev/mL-eq), were calculated from the slope of the regres- Participants were asked to swim at a free pace, resting as much as sion over the linear portion of the dose–response curves. wanted. Before and after swimming, a battery of measurements and biological samples were collected to evaluate exposure and effect bio- 2.2.4. Serum CC16 markers according to a strict schedule (Fig. 1). Timing for biosample Four millilitre of blood were collected before and 1 h after swim- collection was tailored to best capture the aimed biomarkers con- ming and centrifuged at 2500 rpm for 15 min at 4 °C. 100 μL of serum sidering the specific pharmacokinetics, and was based on previous was aliquoted and stored at −80 °C. CC16 was analyzed by ELISA using evidence (Kogevinas et al., 2010; Font-Ribera et al., 2010; Cardador commercial kits (Biovendor Laboratorní medicína a.s., Brno, the Czech and Gallego, 2011; Abramsson-Zetterberg et al., 2000). Blood samples Republic). Intra- and inter-assay coefficients of variation were 3.19% were obtained by venipuncture. Biological samples and measurements and 5.63% respectively. The minimum detectable concentration in at baseline (before swimming) and within 30 min after swimming (ex- serum was set at 46 pg/mL. haled THMs, urine sample) were obtained in a room inside the sports centre but separated from the swimming pool area. Afterwards, parti- 2.3. Biomarkers of exposure cipants were taken by taxi to the Clinical Trial Unit in the Research Centre in a 10 min drive to collect the remaining samples, including 2.3.1. Trihalomethanes in exhaled breath blood samples at 1 h, 24 h and 4 days after swimming and urine samples Trihalomethanes (THMs) including chloroform (CHCl3), bromodi- at 2 h after swimming (Fig. 1). The study was approved by the ethics chloromethane (BDCM), dibromochloromethane (DBCM), and bromo- committee of the research centre following international regulations, form (CHBr3) were measured in exhaled breath before and right after and all volunteers signed an informed consent before participation. swimming. The volunteers breathed deeply and blew three times into the Bio-VOC™ sampler (150 mL), retaining the air for 10 s between each 2.2. Effect biomarkers inspiration and expiration. The alveolar air was collected in a last ad- ditional breath/blow cycle. A screw-in plunger was used to steadily 2.2.1. Micronuclei in peripheral blood lymphocytes (MN-PBL) discharge the sample contained in the Bio-VOC™ into a sorbent tube Four millilitre of blood were obtained with heparinized vacutainers trap. These stainless steel tubes were 3.5″ (89 mm) long and 1/4″ o.d. before and 1 h after swimming. To determine the frequency of binucleated They were packed with 200 mg sorbent Tenax TA 35/60 mesh (Markes cells with MN and the total number of MN, PBLs were cultivated using International Ltd., UK) which was preconditioned using helium of 5 N cytochalasin-B. Slides were stained with 4′-6-diamino-2-fenilindol (DAPI) grade at 100 mL/min at 320 °C for 2 h and then at 335 °C for 30 min. In a fluorescent marker with strong affinity for DNA sequences (Fenech, posterior conditioning cycles they were reconditioned at 335 °C for 2007; Thomas et al., 2009). Each slide was dyed with 15 μL of DAPI, at a 20 min and the same flow carrier gas. One cleaned, the tubes were final concentration of 1 μg/mL, plus Antifade (for preservation of the capped with brass storage caps fitted with PTFE ferrules and were fluorescent signal), and scored immediately or few days after dying at stored at 4 °C in a clean environmental free-solvent atmosphere. x100 with a fluorescent microscope. Samples were stored in the dark at The THMs retained in the tubes were then desorbed and concentrated 4 °C. Five hundred binucleated cells with well-preserved cytoplasm were in a thermal desorption unit equipped with a Unity Series 2 Thermal scored in two replicates by two trained scorers (total of 1000 cells per Desorber and an Ultra 50:50 Multi-tube Auto-sampler (Markes sample) and an average of the two scores was counted per sample. As a International Ltd.). They were desorbed at 300 °C for 5 min and the vapors quality control, each scorer evaluated blindly six times a slide and in all were concentrated into a cold trap at −20 °C at 40 mL/min carrier flow instances except one, differences within scorer was not higher than one (Helium 5 N grade) without split. The cold trap, packed with a graphitized binucleated MN. Differences between scorers for the 230 samples eval- carbon sorbent for General Purpose by Markes International Ltd., was uated were as follows: 21.7% (n = 50) had identical score between the rapidly heated to 300 °C for 5 min with a 7.5 mL/min flow of helium and two scorers; 58.3% (n = 134) differed by one or two binucleated MN; the THMs were transferred to the gas chromatograph (GC) with a split 12.2% (n = 28) differed by 3; 7.8% (n = 18) differed by 4 or more bi- flow of 6 mL/min. Only 20% of the samples in the sorbent tube was in- nucleated MN. These 18 slides with differences above 3 binucleated MN troduced into the GC column by heating at 200 °C. The THMs were finally between scorers were cross-counted and it was found that differences were measured in a Gas Chromatograph 7890 (Agilent Technologies) equipped due to the replica per se and not to differences in criteria between scorers. with a DB-5MS UI capillary column (60 m × 0.32 mm, 1 m; Agilent J &W GC Columns) at 1.5 mL/ml of carrier gas flow (Helium, 5 N grade). The GC 2.2.2. Micronuclei in reticulocytes (MN-Ret) oven temperature program was hold for 10 min at 40 °C; 5 °C/min to Two millilitre of blood was obtained with heparinized vacutainers 150 °C; hold for 10 min; 15 °C/min to 210 °C; and finally hold for 10 min. before and four days after swimming. Samples were stored at room This instrument was coupled to a mass spectrometer (5975C Inert XL temperature until processed within the collection day. Immunomagnetic MSD) with a source in electronic impact mode (Agilent Technologies). The isolation was done to retain transferrine positive immature reticulocytes GC–MS heat transfer line was heated at 280 °C. The MS source and MS from blood (Abramsson-Zetterberg et al., 2000). These immature re- quadrupol temperatures were 230 °C and 150 °C, respectively. The quan- ticulocytes are the very youngest erythrocytes, recently extruded from titative determination was carried out in selected ion monitoring Mode. the bone marrow into the peripheral blood system. After fixation, sam- Dwell time was set at 50 ms for each ion scanned. THMs concentrations ples were stored at 21 °C for three to four days and then at 4 °C. MN were expressed as micrograms per cubic meter (μg/m3). frequency was counted automatically by flow-cytometry in all the iso- lated cells per sample. The assay was performed in the last half of the 2.3.2. Trichloroacetic acid (TCAA) in urine study sample (n = 54). From the 107 samples collected, 56 (52%) had Ten millilitre of urine were collected before and 30 min after swim- detectable levels of isolated cells, with a mean of 27,251 cells/sample ming, stored at −20 °C and shipped in dry ice to the laboratory. TCAA 3 L. Font-Ribera, et al. Environment International 131 (2019) 104988 concentrations were measured using solid phase extraction followed by equilibrium and minor allele frequencies were similar to those de- liquid chromatography tandem mass spectrometry (LC-MS-MS). scribed in the International HapMap Project for European individuals Creatinine in urine was also measured in order to control for excretion (International HapMap Consortium, 2003). rate, and TCAA concentration was expressed as creatinine-adjusted levels (μmol TCAA/mol creatinine). Details on sampling collection and analysis 2.7. Statistical analysis have been published previously (Font-Ribera et al., 2016). Mean or median values were reported to describe central tendencies. 2.4. Environmental measurements 9%, 3%, 19% and 4% of exhaled breath samples before swimming were below the detection limit for chloroform (0.108 μg/m3), bromodi- Chlorine, chloramines, pH, and four THMs were measured in water chloromethane (0.009 μg/m3), dibromochloromethane (0.004 μg/m3) for each of the participants. Nine haloacetic acids (HAAs) in water and bromoform (0.007 μg/m3), respectively, and were imputed half the samples were measured once per experiment day. Water samples were detection limit value. The concentrations of bromodichloromethane, di- composite samples collected at two different points of the pool while bromochloromethane and bromoform were summed up and reported as participants were swimming, stored at 4 °C until laboratory analyses. brominated THMs (Br-THMs). Differences between measures before and Quenching agents in water samples were used for THMs (Na2S2O3_5 after swimming were tested using Wilcoxon signed-rank test. The change H2O and KHSO4) and HAAs (NH4Cl) measurements. Air samples to in the level of exposure and effect biomarkers after swimming was cal- measure trichloramine were collected once per day using a pump with a culated (concentration after–before and mutagenic potency (slope) constant sampling flow rate (1.2 L/min for a mean ± SD of after–before in the case of urine mutagenicity). The distribution of short- 115 ± 32 min) within 1 m from the water and at a height of 60 cm term effect biomarkers was evaluated with a test for normality evaluating above the floor level. The instrumental analyses were performed at the skewness and kurtosis. MN-PBL change showed a normal distribution Institute for Risk Assessment Sciences at Utrecht University (Utrecht, while MN-Ret, urine mutagenicity and CC16 change showed skewed The Netherlands), following the method described by Hery et al. distributions. Spearman correlation was used to estimate the association (1995). Details on sample collection and chemical analysis have been between the change in the effect biomarkers with exposure to DBPs and previously described (Font-Ribera et al., 2016). exercise. Potential confounding was evaluated among individual cov- ariables including age, sex, body mass index, current swimming practice, 2.5. Physical activity and personal information distance swam, time swimming, energy expenditure, intensity of ex- ercise, exertion, and shortness of breath. Confounders were identified A study field worker sitting in front of the pool counted the distance empirically as those covariables related both with the effect biomarker swam and the time swimming by each participant with a counter and a and exposure. Only distance swam was identified, associated with CC16 chronometer. Energy expenditure (in kcal) was estimated using the and DBP exposure. Spearman partial correlation adjusting for distance swimming speed and the weight of the participant, assuming that swam was used for CC16, based on robust linear regression. 95% con- swimming at 46 m/min equals 8.3 metabolic equivalent tasks (METs; fidence intervals of correlation coefficients were estimated through kcal per kg per hour) (Ainsworth et al., 2011). Subjective exertion and bootstrap with 1000 iterations. The p-value threshold for statistical sig- shortness of breath were evaluated before and after swimming using the nificance was set up at < 0.05. Genetic analyses were performed as- Borg scale (Borg, 1970) with the following questions “In a scale from 0 suming a dominant genetic model, considering the most frequent allele to 10, how would you describe your exertion/shortness of breath?”, and as a reference category. To evaluate interactions between changes in the change after swimming was calculated for both parameters. Heart THMs and gene variants, we included an interaction term for dichot- rate during the 40 min was measured using a heart rate monitor (Polar omous genotype and exhaled THMs in linear regression models. All RCX5, Polar Electro Oy, Kempele, Finland), and the intensity of phy- statistical analyses were performed with the statistical package STATA sical activity was calculated in relation to the individual theoretical 12.0 (StataCorp. College Station, TX, USA). maximum heart rate (MHR) (Gulati et al., 2010). The heart rate < 50%, between 50 and 69%, and > 69% of MHR were considered as 3. Results low, moderate, and high intensity, respectively. Weight and height were measured with standard procedures and the body mass index A total of 116 subjects (56 men, 60 women) participated in the (BMI) was calculated. Questionnaires were used to collect information study, with a median age of 23.9 years (Table 1). The majority was on sociodemographic data, health status, lifestyle, frequency of swim- university students and 28% attended swimming pools at least once a ming pool attendance and other physical activities. month. The median distance swam during the study session was 1.1 km and median percentage time at high intensity physical activity was 2.6. Gene selection and genotyping 71%, according to heart rate parameters. There was an increase in the exertion and shortness of breath scores after swimming (3 and 2 points, Blood was collected in EDTA tubes and stored at −80 °C until DNA respectively). Energy expenditure was unrelated to age, while it was extraction. DNA was extracted using QIAamp DNA Blood Mini Kit higher for men than for women (Appendix Table A1). BMI was nega- (QIAGEN) and was measured with NANODROP 8000 (Thermo- tively related to distance swam and positively to energy expenditure Scientific) and all samples were diluted to a fixed concentration of and shortness of breath. 2.5 ng/mL. Twelve polymorphisms in three genes involved in DBP Median concentrations of DBPs in pool water were 37.3 μg/L for metabolism were genotyped: three single-nucleotide polymorphisms chloroform, 0.7 μg/L for bromoform, 48.5 μg/L for total THMs (SNPs) in the glutathione S-transferase zeta 1 (GSTZ1) gene, eight SNPs (TTHMs), and 111.2 μg/L for total HAAs (Table 2). Median level of in the Cytochrome P450 2E1 (CYP2E1) gene by Taqman allele dis- trichloramine in air was 472.6 μg/m3, ranging from 248.5 to 858.3 μg/ crimination assay (Life Technologies) and a common deletion in the m3. Additional information on DBP levels in water and air is shown in glutathione S-transferase theta 1 (GSTT1) gene. For a complete list of Table 2. For technical reasons some of the samples were lost and the SNPs see Appendix Table A2. For quality control purposes, duplicate final observations are shown in Table 2. samples (5% of the total numbers of samples) were repeated for each There was a significant increase after swimming in the level of all SNP, and no template controls (NTCs) were included in each plate. All exposure biomarkers (Table 3). Median increase in exhaled breath after polymorphisms could be successfully analyzed in the 116 participants swimming was 10.9 μg/m3 for chloroform (range 2.1, 35.1), 2.6 μg/m3 (the quality control of genotypes was assured with > 99% of con- for Br-THMs (range 0.1, 8.5) and 13.4 μg/m3 for TTHMs (range 2.6, cordance). All polymorphisms were also in Hardy-Weinberg 42.6). Creatinine-adjusted TCAA in urine increased a median value of 4 L. Font-Ribera, et al. Environment International 131 (2019) 104988 Table 1 dibromochloromethane and bromoform, and urinary trichloroacetic Characteristics of the study population and the exercise performed (N = 116). acid increased after swimming. Among the four evaluated biomarkers Variable Median Percentile 25, 75 of short-term effect (MN-PBL, urine mutagenicity, serum CC16, and MN-Ret in a subset of 19), only change in MN-Ret showed a moderate Age, years 23.9 21.3, 28.9 association with DBP exposure. However, sample size was small and Body mass index, kg/m2 23.0 21.3, 25.5 interpretation of these finding should be cautious. Distance swam, km 1.1 0.8, 1.2 Time swimming, min 35.3 29.9, 38.7 MN-PBL is a well-established biomarker of DNA damage and in Energy expenditure, kcal 204.5 166.9, 254.6 adults has been associated with cancer risk in large longitudinal studies Percentage of time at high intensity (> 69% MHR) 70.6 33.1, 88.2 (Bonassi et al., 2011). It has also been linked to environmental ex- Exertiona 3 2, 4 posures, both at long-term, including Br-THMs in tap water (Stayner Shortness of breatha 2 1, 3.5 et al., 2014), and at short-term, such as ozone (Holland et al., 2015). MHR: maximum heart rate. MN-PBL was determined by manually counting MN in 1000 binucleated a Difference in the Borg scale (0–10 points), measured before and after cells per sample. MN-Ret has been recently proposed as a genotoxicity swimming. biomarker. It is more sensitive (Kissling et al., 2007), using an auto- mated flow cytometer-based analysis, allowing the measurement of a 3.1 μmol/mol (range -9.23, 38.89), and 15 subjects showed higher le- large number of cells (i.e., around 27,000 per sample in the present vels before compared to after swimming. None of the effect biomarkers study). In this method, no cultivation of cells occurs but a disadvantage showed a significant change after swimming (Table 3). is the complex sample preparation needed in order to isolate re- The association between individual characteristics and change in ticulocytes that occur in a very low proportion in blood (Abramsson- biomarkers are shown in Table 4. Men showed a larger increase in Zetterberg et al., 2000). MN in PBL and in reticulocytes reflects dif- exhaled Br-THMs after swimming, while women had a larger increase ferent exposures windows, as the half-life of the cells differs. A corre- in creatinine-adjusted TCAA in urine. Several measures of physical lation between MN-Ret and MN-PBL frequencies has been found in activity were positively related to the increase in THMs in exhaled previous studies both at baseline levels (Costa et al., 2011) and in the breath after swimming, while distance swam and energy expenditure increase after an acute exposure (Stopper et al., 2005). The rather small were negatively related to the increase in creatinine adjusted TCAA in numbers and differential influence of exposures other than those re- urine. The change in the effect biomarkers did not differ by personal lated to swimming, particularly for MN-Ret, complicate the direct in- characteristics, except for serum CC16, whose change was negatively terpretation of the associations. In the present study, positive correla- correlated to BMI and positively correlated to distance swam (Table 4). tions were found among the two assays but these were not statistically The baseline level of serum CC16 was negatively correlated with BMI significant, probably due to the low sample size. and significant differences were also detected by regular physical ac- The present study has been conducted exactly in the same swim- tivity and the polymorphism in the CC16 gene (Appendix Table A2). ming pool of our previous PISCINA study (Kogevinas et al., 2010; Font- The baseline frequency of MN-PBL was positively correlated to age and Ribera et al., 2010). The protocols of the two studies were very similar was higher among ever smokers. Urine mutagenicity before swimming differed among different genotypes of polymorphisms in CYP2E1 and Table 2 GSTZ1 (Appendix Table A2). The frequencies of MN-PBL and MN-Ret Level of disinfection by-products and related parameters in the swimming pool were positively but not significantly correlated at baseline and after water. swimming (Spearman correlations of 0.38 (p-value 0.134) and 0.24 (p- value 0.330), respectively). Parameter N Median Percentile 25, Min Max 75 Table 5 shows the association between the change in short-term effect biomarkers and exposure. No associations were found for MN- Trihalomethanes (THMs) (μg/L) PBL, urine mutagenicity and CC16. A moderate correlation was found Chloroform 108 37.3 32.9, 43.8 24.0 61.6 between MN-Ret and DBP exposure, positive for chloroform (0.50), Bromodichloromethane 108 7.1 5.4, 8.6 3.8 12.9 Dibromochloromethane 108 2.0 1.6, 3.0 0.9 4.7 brominated THMs (0.55) and total THMs (0.56), and negative for urine Bromoform 108 0.7 0.5, 1.3 0.2 1.9 TCAA (−0.59) (Table 5). Brominated THMs 108 9.5 7.9, 12.6 5.7 19.4 We examined interactions between Br-THMs and TTHMs and Total THMs 108 48.5 43.6, 54.7 30.7 74.7 polymorphisms in GSTT1, GSTZ1, and CYP2E1 metabolism genes in Haloacetic acids (HAAs) (μg/L) relation to frequency of MN-PBL (Appendix Table A3). Statistically Dichloroacetic 30 29.4 22.3, 37.7 15.4 51.7 Trichloroacetic 30 62.9 57.4, 70.2 39.2 83.4 significant interactions were observed between five SNPs in the CYP2E1 Bromochloroacetic 30 4.9 4.2, 6.2 2.4 8.8 gene, exposure to THMs and MN-PBL frequency; interactions for two Dibromoacetic 30 1.0 1.0, 1.8 0.5 3.1 SNPs (rs2249695 and rs2515641) remained statistically significant Dichlorobromoacetic 30 11.8 9.4, 15.4 4.8 23.4 even after Bonferroni correction for multiple comparisons. There was Total HAAs 30 111.2 95.5, 129.3 73.3 144.3 no difference in the association between THM exposure and MN-PBL Chlorine (mg/L as Cl2) Free chlorine 114 1.3 1.0, 1.6 0.6 2.0 frequency, between the null GSTT1 genotype (−/−; a deletion in both Total chlorine 114 2.2 1.9, 2.3 1.6 2.7 copies of the gene) and those with one (+/−) or none (+/+) of the Chloramines (mg/L as Cl2) copies deleted. There was also no interaction for three SNPs of the Monochloramine 114 0.2 0.1, 0.4 0.0 0.7 GSTZ1 gene. We also examined the rs3741240 polymorphism of the Dichloramine 112 0.3 0.0, 0.4 0.0 0.7 Trichloramine 114 < DL < DL, 0.6 < DL 1.6 CC16 gene in relation to levels of the CC16 protein and found no in- Other parameters teraction with levels of CC16 protein. Trichloramine in air (μg/m3) 26 472.6 381.4, 533.5 248.5 858.3 pH 115 7.40 7.3, 7.6 7.1 8.0 Water temperature (°C) 115 28.0 27.8, 28.2 27.1 28.6 4. Discussion Air temperature (°C) 115 28.5 27.8, 29 26.5 29.5 Number of swimmers in the 115 23.0 16.3, 34.3 9.0 120.5 pool We evaluated DBP exposure and markers of genotoxicity and lung epithelium permeability in 116 young subjects after swimming for All samples were below the detection limit (DL) for chloroacetic acid (< 3 μg/ 40 min in an indoor chlorinated pool. Personal exposure to disinfection L), bromoacetic acid (< 2 μg/L), dibromochloroacetic acid (< 2 μg/L), and by-products measured as exhaled chloroform, bromodichloromethane, tribromoacetic acid (< 5 μg/L). 5 L. Font-Ribera, et al. Environment International 131 (2019) 104988 Table 3 Levels of the biomarkers of exposure to disinfection by-products and genotoxicity, mutagenicity and lung epithelium permeability. Biomarker N Before swimming After swimming Change p-Valuea Time after swim Median Median Median (pctile25, 75) (pctile25, 75) (pctile25, 75) DBP exposure Chloroform in exhaled breath, μg/m3 116 0.4 (0.2, 0.6) 11.5 (8.0, 14.7) 10.9 (7.6, 14.3) < 0.001 5 min Br-THMs in exhaled breath, μg/m3 116 0.1 (0.0, 0.1) 2.7 (2.0, 4.1) 2.6 (1.9, 3.9) < 0.001 5 min TTHMs in exhaled breath, μg/m3 116 0.5 (0.3, 0.7) 14.4 (10.0, 18.4) 13.4 (9.7, 17.7) < 0.001 5 min TCAA in urineb, μmol/mol 113 2.5 (1.5, 4.8) 5.8 (3.2, 11.2) 3.1 (1.1, 6.1) < 0.001 30 min Effect MN-PBL (‰) 113 4 (3,6) 4 (3,7) 0 (−2, 2) 0.185 1h MN-Ret (‰) 19 0.9 (0.6, 1.3) 1.1 (0.8, 2.3) 0.4 (−0.4, 0.8) 0.277 4 days Urine mutagenicity (rev/mL-eq) 88 1.3 (0.4, 2.0) 1.1 (0.4, 1.8) −0.1 (−0.8, 0.5) 0.434 2h Serum CC16 (ng/mL) 105 11.1 (8.7, 13.4) 10.5 (8.2, 13.2) −0.5 (−1.2, 1.2) 0.289 1h Br-THMs: brominated trihalomethanes; TTHMs: total trihalomethanes; TCAA: trichloroacetic acid. MN-PBL: micronuclei in peripheral blood lymphocytes. MN-Ret: micronuclei in reticulocytes. a Wilcoxon test. b Creatinine adjusted. in study design, population characteristics, timing of sample collection quality in the city (ASPB, 2012). Given that chloroform is a non-gen- and analytical techniques. Although median level of total THMs are otoxic THM, this could explain the lack of consistent effects on the similar in both studies, 45.5 μg/L in PISCINA (Font-Ribera et al., 2010) genotoxicity and mutagenicity biomarkers in the present study, while compared to 48.5 μg/L in the present study, the composition has the previous PISCINA study showed associations between MN-PBLs and completely changed. Brominated THMs were the most predominant urine mutagenicity with exhaled Br-THMs levels, that were not found THMs in the previous study, while chloroform is the predominant THM for chloroform (Kogevinas et al., 2010). The results for MN-Ret require in the present. Median chloroform level was 16.1 μg/L in the previous replication with a larger sample size. In particular, the association be- study (35% of total THMs) and 37.3 μg/L in the present (77% of total tween MN-Ret and chloroform is difficult to understand since chloro- THMs). This difference is also observed in exhaled levels, and although form itself is not considered genotoxic, although chloroform could be total THMs in exhaled breath in the present study is 14.4 compared to an indicator of other DBPs. However, chance cannot be ruled out given 7.9 μg/m3 in the previous study, the levels of brominated THMs in the limited sample size. exhaled breath have declined (from 0.5 to 0.1 μg/m3 in exhaled bro- As shown in Table 4, there are significant correlations between THM moform levels). This decrease reflects mostly changes in the tap water concentrations in exhaled breath and distance swam or energy Table 4 Spearman correlation coefficients between individual characteristics and the changes in the exposure and effect biomarkers, and median values by sex and current swimming. Covariables Exposure biomarkers Effect biomarkers Chloroform in Br-THMs in TTHMs in Creatinine adjusted MN-PBLs MN-Ret Urine mutagenicity CC16 in exhaled breath (μg/ exhaled breath exhaled breath TCAA in urine (μmol/ (‰) (‰) (rev/mL-eq) serum (ng/ m3) (μg/m3) (μg/m3) mol) mL) N 116 116 116 113 113 19 88 105 Age −0.08 0.04 −0.05 0.13 −0.12 −0.14 −0.01 −0.14 Body mass index 0.06 0.07 0.07 −0.05 0.07 −0.13 −0.03 −0.25⁎ Distance swam, km 0.26⁎⁎ 0.21⁎ 0.25⁎⁎ −0.32⁎⁎ −0.06 −0.01 −0.18 0.25⁎ Time swimming, min 0.05 0.14 0.07 −0.21⁎ −0.01 −0.23 −0.11 0.11 Energy expenditure, 0.27⁎⁎ 0.24⁎⁎ 0.27⁎⁎ −0.33⁎⁎ 0.00 0.21 −0.14 0.19 kcal High intensitya 0.16 0.21⁎ 0.19⁎ −0.14 −0.06 0.02 −0.06 0.18 Exertionb 0.22⁎ 0.37⁎⁎ 0.26⁎⁎ 0.03 −0.06 0.23 −0.09 0.01 Shortness of breathb 0.21⁎ 0.34⁎⁎ 0.24⁎⁎ 0.03 0.04 0.37 0.02 −0.11 Sex Males, median 11.4 3.0⁎ 15.0 1.8⁎⁎ 1 0.44 −0.11 0.01 Females, median 10.7 2.5⁎ 12.7 4.0⁎⁎ 0 −0.47 −0.11 −0.66 Current swimmingc Yes, median 13.5⁎⁎ 3.2 16.7⁎ 2.0 0 0.44 −0.12 −0.52 No, median 10.6⁎⁎ 2.6 13.0⁎ 3.6 0.5 0.05 −0.11 −0.45 Br-THMs: brominated trihalomethanes; TTHMs: total trihalomethanes; TCAA: trichloroacetic acid. MN-PBL: micronuclei in peripheral blood lymphocytes. MN-Ret: micronuclei in reticulocytes. p-Values from the Spearman correlations, except for sex and current swimming (Kolmogorov-Smirnov test). ⁎ p-Value < 0.05. ⁎⁎ p-Value < 0.01. a > 69% maximum heart rate. b Difference in the Borg scale (0–10 points), measured before and after swimming. c At least once per month. 6 L. Font-Ribera, et al. Environment International 131 (2019) 104988 Table 5 Spearman correlation coefficients (95% confidence interval) between the change in short-term effect biomarkers and exposure to disinfection by-products and exercise. Exposure MN-PBL N MN-Ret N Urine mutagenicity N CC16a N (‰) (‰) (rev/mL-eq) (ng/mL) Increase in exhaled breath Chloroform, (μg/m3) −0.16 113 0.50 19 −0.13 88 −0.01 105 (−0.35, 0.04) (0.07, 0.93) (−0.33, 0.08) (−0.20, 0.18) 3 Brominated THMs, (μg/m ) −0.05 113 0.55 19 −0.14 88 −0.05 105 (−0.25, 0.14) (0.13, 0.97) (−0.36, 0.07) (−0.25, 0.14) 3 Total THMs, (μg/m ) −0.15 113 0.56 19 −0.14 88 −0.01 105 (−0.34, 0.05) (0.14, 0.97) (−0.36, 0.07) (−0.21, 0.18) Increase in TCAA in urine, (μmol/mol) −0.02 110 −0.59 18 −0.06 86 −0.09 102 (−0.21, 0.17) (−0.94, −0.24) (−0.27, 0.16) (−0.30, 0.12) Trichloramine in air, (μg/m3) 0.05 99 0.20 17 −0.002 77 0.03 91 (−0.14, 0.24) (−0.37, 0.77) (−0.23, 0.23) (−0.18, 0.25) Distance swam, km −0.06 113 −0.01 19 −0.18 88 (−0.25, 0.14) (−0.47, 0.45) (−0.39, 0.04) Energy expenditure, kcal 0.0003 113 0.21 19 −0.14 88 (−0.19, 0.18) (−0.25, 0.67) (−0.37, 0.09) High intensity (> 69% MHR) 0.06 109 0.92 19 −0.06 85 (−0.25, 0.12) (−0.45, 0.50) (−0.29, 0.17) MHR: maximum heart rate; MN-PBL: micronuclei in peripheral blood lymphocytes; MN-Ret: micronuclei in reticulocytes; TCAA: trichloroacetic acid. a Spearman's partial correlation coefficient adjusted for distance swam. expenditure, p < 0.05 or 0.01. These observations are consistent with adjustment for physical activity may suggest that the intensity of phy- inhalation as primary mechanism for THM incorporation into swim- sical activity was not sufficient to trigger this change. mers (Marco et al., 2015). However, the air-water partition mechanism We found interactions between some polymorphisms in the CYP2E1 of the pulmonary system leads to higher retention of the THM with gene, exposure to brominated THMs and MN-PBL. Metabolism of DBPs lower Henry's Law volatility constants, e.g. those with higher degree of is mediated by enzymes from the glutathione S-transferase (GST) and bromination (Marco et al., 2015) and therefore a selective relative cytochrome P450 (CYP) families. Polymorphisms in GSTT1, GSTZ1 and enrichment of these compounds in the exhaled breath. Given that the CYP2E1 have been suggested to modify DBP-associated bladder cancer exposure event occurred in a short time frame (10:00 to 11:42) in all risk in a case-control study (Cantor et al., 2010) and to affect the DBP volunteers, circadian variation of activity in genes involved in the internal dose after swimming (Font-Ribera et al., 2016). Findings on metabolism of DBPs (Gängler et al., 2018) are expected to be minimal. gene-environment interactions that remained significant after cor- Short-term changes in serum CC16 may reflect an increase in lung recting for multiple comparisons add to the global evidence on the epithelium permeability (Broeckaert et al., 2000), and it has been used importance of genetic polymorphisms in mediating the effect of disin- as a sensitive biomarker to detect respiratory effects after short-term fection by-products in water. Still the study had limited power to exposures such as air pollution (Stockfelt et al., 2012; Provost et al., evaluate gene–environment interactions and we cannot exclude the 2014). The PISCINA2 study is the first to evaluate the relationship possibility of having spurious findings. between the change in serum CC16 after swimming and the level of The timing of the collection of biological samples is crucial when trichloramine in air, the main potential airway irritant in swimming evaluating biomarkers of effect. For PBL-MN, CC16 and urine geno- pools. A large variability in trichloramine levels was observed during toxicity we followed the same timing as in our previous PSCINA study, the thirty experimental days (from 248.5 to 858.3 μg/m3), with a where changes in those biomarkers were observed (Font-Ribera et al., median level comparable or even higher than in previous studies 2010; Kogevinas et al., 2010). For other biomarkers (MN in re- (Carbonnelle et al., 2002; Carbonnelle et al., 2008; Font-Ribera et al., ticulocytes, urinary TCAA) we considered previous knowledge and ex- 2010; Nordberg et al., 2012). The lack of association detected between pert judgement of researchers to establish the times to collect specific trichloramine in air and serum CC16 suggests that DBP exposure in a samples to measure specific biomarkers. single swimming session at the observed levels is not enough to affect The main advantage of the pre-post design is that individuals are this respiratory biomarker. compared with themselves, which is crucial for assessing small changes in In the previous PISCINA study we detected a 3% increase in serum biomarkers that may have a large inter-individual variability. A potential CC16 1 h after swimming (Font-Ribera et al., 2010), that was associated limitation is that external factors, such as physical activity, may still affect with DBPs exposure measured as THMs in exhaled air, and to physical results even after adjustment. Alternative designs have been proposed such exercise assessed through self-reported distance swam. This suggested a as swimming pools that use different disinfectants (Carbonnelle et al., potential confounding effect by physical activity. In the current study 2002; Fernández-Luna et al., 2012; Llana-Belloch et al., 2015) or repeat we improved the measurement of physical activity by including dif- the measurements after physical exercise without DBP exposure, such as ferent objective and subjective indicators. The distance swam, counted indoor cycling or running. DBPs are, however, present in all disinfected by a technician, was the measure that correlated best to changes in pools but in a different composition and in our pool there was considerable serum CC16 and also to the DBP exposure biomarkers. A previous be- variability in DBPs levels in different days. Other sports may not be fore-after study detected a 44% and 52% increase in CC16 among 14 comparable concerning biological changes due to physical activity. In this trained swimmers after swimming 45 min over a distance of 1500 m in context, the PISCINA2 study followed a pre-post design with an accurate a chlorinated and in a non-chlorinated swimming pool, respectively individual measurement of DBPs and physical activity, and the different Carbonnelle et al. (2002). However, they did not find change in CC16 factors were analyzed in multivariate models. A proper measurement of among 13 adults under a recreational regime during 2 h (1 h standing physical activity is especially relevant in the assessment of respiratory on the pool side, 1 h swimming freely). The lack of increase in serum effects, but may be less crucial when assessing genotoxicity biomarkers. CC16 despite a similar exercise compared to the previous study (median We focused the exposure assessment on exhaled THMs, urinary distance swam 1.1 km here vs. 0.95 km in PISCINA1) and the better TCAA and trichloramine in air. However, urinary TCAA has not been 7 L. Font-Ribera, et al. Environment International 131 (2019) 104988 validated as an exposure marker in short-term exposure settings like metabolomics (van Veldhoven et al., 2018) and transcriptomics (Espín- before-after swimming, which may partly explain the negative asso- Pérez et al., 2017) reported separately. ciations with Ret-MN. Although we measured other DBPs in water (Font-Ribera et al., 2016), we focused our main analysis in the bio- 5. Conclusions marker-based exposure measures that provide more accurate personal exposure. Alternative analyses using DBPs in water as exposure did not The unchanged levels of the short-term effect biomarkers after show different results from the ones presented here. Concerning CC16, swimming and null associations with personal estimates of exposure to despite the focus of our study was on trichloramine, other DBPs present disinfection by-products suggest no measurable effect on genotoxicity in air could contribute to lung irritation (e.g. dichloroacetonitrile, cy- in lymphocytes, urine mutagenicity and lung epithelium permeability anogen chloride). at the observed exposure levels. The moderate associations with mi- This is the largest study on the topic with 116 subjects evaluated. cronuclei in reticulocytes require cautious interpretation given the The exposure assessment to DBPs is also the most complete, with dif- limited sample size. ferent classes of DBPs measured every day including not only THMs, but more carcinogenic DBPs such as HAAs, haloacetonitriles or ni- Declaration of Competing Interest trosamines (Plewa et al., 2008), that have been previously published (Font-Ribera et al., 2016). TCAA in urine was used to complement ex- Authors have no competing interests. haled THMs as an exposure biomarker, although it was poorly corre- lated to other DBPs in water (Font-Ribera et al., 2016). Details on the Acknowledgements levels, correlations and determinants of the evaluated exposure bio- markers have been described elsewhere (Font-Ribera et al., 2016). We are grateful to all volunteers and the swimming pool facility. Timing of sample collection was based on previous studies (Abramsson- This work was funded by the EU 7th Framework Programme Zetterberg et al., 2000; Kogevinas et al., 2010; Font-Ribera et al., 2010; EXPOSOMICS Project (grant agreement 308610). We thank the Parc de Cardador and Gallego, 2011), although evidence on the expression Salut Mar Biobank (MARBiobanc) for providing support to the project. dynamics of the biomarkers after an environmental exposure is scarce. MARBiobanc is supported by grants from Instituto de Salud Carlos III/ The DBP levels in the study swimming pool are similar to those reported FEDER (RD09/0076/00036 and PT13/0010/0005) and the “Xarxa de in other studies (Chowdhury et al., 2014), supporting the external va- Bancs de tumors” sponsored by Pla Director d'Oncologia de Catalunya lidity of these results. Another novelty of the PISCINA2 study is that we (XBTC). ISGlobal is a member of the CERCA Programme, Generalitat de measured for the first time the effect of swimming in several omic Catalunya. We appreciate the work by Kate Jones (Health & Safety signatures including proteomics (Vlaanderen et al., 2017), Laboratory, UK) for urinary TCAA analysis. Appendix A Appendix Table A1 Distribution of physical activity indicators and associations with sex, age and body mass index. Variable All Men Women Age Body mass index Median Median Median Spearman correlation coefficient Spearman correlation coefficient (pctile 25, 75) (pctile 25, 75) (pctile 25, 75) Distance swam, km 1.0 1.1 1.0 −0.11 −0.20⁎ (0.8, 1.2) (0.8, 1.2) (0.9, 1.2) Energy expenditure, kcal 204.5 228.8 197.5 −0.02 0.26⁎⁎ (166.9, 254.6) (175.5, 310.2)⁎ (161.4, 233.8)⁎ Time swimming, min 35.3 35.3 35.3 −0.07 −0.13 (29.8, 38.7) (27.9, 38.9) (31.4, 37.9) High intensity (> 69% MHR), time percentage 70.7 63.6 78.3 −0.14 −0.10 (33.1, 88.2) (35.7, 84.4) (33.1, 89.2) Exertion, Borg scale (0−10) Before 1 (0, 2) 1 (0, 1.5) 1 (0, 2) 0.04 0.04 After 4 (3, 5) 4.5 (4, 6)⁎ 3.7 (3, 5)⁎ −0.09 0.16 Difference 3 (2, 4) 3 (2.5, 5) 3 (2, 4) −0.11 0.11 Shortness of breath, Borg scale (0–10) Before 0 (0, 1) 0 (0, 1) 0 (0, 1.5) −0.15 −0.10 After 3 (1, 4) 2.5 (1, 5) 3 (1.5, 4) −0.04 0.14 Difference 2 (1, 3.5) 2 (0, 4) 2 (1, 3) 0.00 0.21⁎ MHR: maximum heart rate. P-values from Kolmogorov-Smirnov test for sex and from Spearman correlations for age and body mass index. ⁎ p-Value < 0.05. ⁎⁎ p-Value < 0.01. Appendix Table A2 Potential determinants of the baseline levels of the genotoxicity and respiratory biomarkers. Variable Category MN-PBL MN-Ret Urine mutagenicity Serum CC16 N Spearman's corr N Spearman's corr N Spearman's corr N Spearman's corr Age (years) 113 0.34⁎⁎ 19 −0.05 97 0.00 108 0.08 Body mass index (kg/m2) 113 −0.07 19 0.05 97 0.01 108 −0.22⁎ (continued on next page) 8 L. Font-Ribera, et al. Environment International 131 (2019) 104988 Appendix Table A2 (continued) Variable Category MN-PBL MN-Ret Urine mutagenicity Serum CC16 N Spearman's corr N Spearman's corr N Spearman's corr N Spearman's corr N Median N Median N Median N Median (pc25, pc75) (pc25, pc75) (pc25, pc75) (pc25, pc75) Sex Male 56 4 (3, 5) 13 0.9 (0.7, 1.1) 47 1.2 (0.4, 2.1) 51 11.5 (9.1, 14.0) Female 60 5 (3, 7) 6 1.2 (0.6, 1.6) 50 1.0 (0.3, 1.6) 57 10.7 (7.9, 12.0) Ever smoker Yes 16 5 (4, 6.5)⁎ 4 0.6 (0.4, 0.8) 16 0.9 (0.1, 1.4) 16 11.3 (9.2, 13.9) No 96 4 (3, 6)⁎ 15 1.1 (0.8, 1.4) 81 1.3 (0.4, 1.9) 91 11.0 (8.2, 13.4) Drinking water type Bottled 52 5 (3, 6) 6 0.9 (0.6, 1.1) 45 1.0 (0.3, 1.9) 49 11.1 (8.7, 12.7) Tap filtered 21 4 (2, 8) 5 1.3 (0.9, 1.6) 15 1.3 (0.3, 1.5) 21 10.6 (9.2, 13.6) Tap unfiltered 32 4 (3, 6) 8 0.9 (0.6, 1.2) 30 1.1 (0.2, 1.5) 30 11.0 (7.4, 13.8) Current swimminga Yes 33 4 (3, 6) 9 1.1 (0.9, 1.3) 29 1.1 (0.1, 1.7) 32 11.0 (7.7, 13.5) No 80 5 (3, 6) 10 0.8 (0.5, 1.1) 68 1.1 (0.4, 1.9) 76 11.1 (8.7, 13.2) Physical activityb Low 53 5 (3, 6) 7 0.7 (0.3, 1.1) 49 1.2 (0.3, 1.7) 50 11.0 (9.2, 12.9) Medium 34 4.5 (2, 6) 5 1.0 (0.8, 1.3) 27 1.1 (0.2, 2.1) 33 9.1 (7.6, 12.4) High 25 4 (3, 6) 7 1.1 (0.6, 1.4) 20 1.1 (0.4, 1.7) 24 12.8 (10.8, 15.2)⁎ Asthma Yes 11 4 (2, 5) 3 0.6 (0.5, 1.1) 10 0.4 (0.3, 1.0) 9 9.3 (8.8, 10.5) No 101 5 (3, 6) 16 1.0 (0.8, 1.3) 87 1.3 (0.4, 1.9) 98 11.1 (8.6, 13.6) GSTT1 (glutathione S-transferase theta 1 gene) Null 23 5 (2, 6) 2 1.6 (1.4, 1.8) 21 1.5 (1.0, 1.8) Positive 90 4 (3, 6) 17 0.9 (0.6, 1.1) 76 0.9 (0.3, 1.9) CYP2E1 (Cytochrome P450 2E1 gene) rs2249695 CC 77 4 (3, 6) 15 1.0 (0.7, 1.3) 67 1.3 (0.3, 2.0) CT/TT 36 5 (2.5, 6) 4 0.7 (0.4, 1.2) 30 0.8 (0.3, 1.5) rs8192766 TT 89 4 (3, 6) 17 0.9 (0.6, 1.1) 75 1.3 (0.4, 1.9) TG/GG 24 4.5 (3, 6) 2 1.4 (1.3, 1.6) 22 0.5 (0.3, 1.3) rs915906 TT 85 4 (3, 6) 15 1.0 (0.7, 1.3) 74 1.3 (0.4, 2.0) TC/CC 28 4.5 (2, 6.5) 4 0.7 (0.4, 1.2) 23 0.8 (0.2, 1.4) rs915907 CC 86 4.5 (3, 6) 16 0.9 (0.6, 1.2) 73 1.3 (0.4, 2.0) CA/AA 27 4 (3, 6) 3 1.3 (0.8, 1.6) 24 0.5 (0.3, 1.4) rs2070673 TT 76 4 (3, 6) 15 1.0 (0.7, 1.3) 65 1.3 (0.4, 1.9) TA/AA 37 5 (3, 6) 4 0.9 (0.4, 1.4) 32 0.7 (0.2, 1.4) rs2070676 CC 86 4 (3, 6) 16 1.0 (0.8, 1.3) 74 1.3 (0.4, 1.9) CG/GG 27 5 (2, 7) 3 0.6 (0.3, 1.6) 23 0.9 (0.3, 1.5) rs2031920 CC 104 4 (3, 6) 18 0.9 (0.6, 1.3) 89 1.3 (0.4, 1.9)⁎ CT/TT 9 5 (4, 6) 1 1.3 8 0.4 (0.1, 0.6) rs2515641 CC 87 4 (3, 6) 16 1.0 (0.8, 1.3) 75 1.2 (0.3, 1.9) CT/TT 26 5 (3, 7) 3 0.6 (0.3, 1.6) 22 1.0 (0.3, 2.3) GSTZ1 (glutathione S-transferase zeta 1 gene) rs1046428 CC 67 4 (3, 6) 11 0.9 (0.6, 1.1) 57 0.8 (0.3, 1.7)⁎ CT/TT 46 4.5 (3, 6) 8 1.1 (0.6, 1.5) 40 1.4 (0.8, 1.9) rs7972 GG 95 4 (3, 6) 17 0.9 (0.6, 1.3) 81 1.2 (0.4, 1.9) GA/AA 18 4.5 (2, 5) 2 6.7 (1.0, 12.4) 16 1.1 (0.3, 1.7) rs7975 GG 48 4 (3, 6) 10 0.8 (0.6, 1.1) 42 1.0 (0.3, 2.0) GA/AA 65 5 (3, 7) 9 1.1 (0.9, 1.3) 55 1.2 (0.4, 1.8) CC16 (club cell protein gene) rs3741240 GG 54 11.6 (9.3, 13.3)⁎ GA/AA 54 9.5 (7.9, 13.4) MN-PBL: Micronuclei in peripheral blood lymphocytes. MN-Ret: Micronuclei in reticulocytes. a At least once per month. b Days per week doing vigorous physical activity during leisure time (low: 0, medium: 1–2, high: > 2). ⁎ p-Value < 0.05, calculated using Spearman correlation for continuous variables, Kolmogorov-Smirnov for dichotomous variables and Kruskall-Wallis for ca- tegorical variables with three categories. Appendix Table A3 Effect modification by single nucleotide polymorphisms of the association between the change in micronuclei frequency in peripheral blood lymphocytes (MN-PBL) for an interquartile range increase in total (TTHM) and brominated (Br-THM) trihalomethanes in exhaled breath after swimming. Beta coefficients from linear regression. Gene/polymorphism Variant N TTHM Br-THM β Coef. (95%CI) β Coef. (95%CI) 113 −0.47 (−1.02, 0.08) −0.14 (−0.77, 0.49) GSTT1 (glutathione S-transferase theta 1) Null 23 −0.66 (−1.75, 0.42) −0.38 (−1.62, 0.86) Positive 90 −0.44 (−1.08, 0.20) −0.10 (−0.83, 0.63) Interaction p-value 0.791 0.765 CYP2E1 (cytochrome P450 2E1) rs2249695 CC 77 −0.95 (−1.58, −0.33) −0.87 (−1.61, −0.13) CT/TT 36 0.71 (−0.39, 1.82) 1.28 (0.19, 2.37) (continued on next page) 9 L. Font-Ribera, et al. Environment International 131 (2019) 104988 Appendix Table A3 (continued) Gene/polymorphism Variant N TTHM Br-THM β Coef. (95%CI) β Coef. (95%CI) Interaction p-value 0.007 0.001⁎ rs8192766 TT 89 −0.52 (−1.12, 0.07) −0.24 (−0.93, 0.45) TG/GG 24 −0.34 (−2.16, 1.47) 0.30 (−1.58, 2.17) Interaction p-value 0.837 0.560 rs915906 TT 85 −0.94 (−1.58, −0.29) −0.69 (−1.46, 0.07) TC/CC 28 0.67 (−0.36, 1.70) 0.92 (−0.13, 1.97) Interaction p-value 0.010 0.018 rs915907 CC 86 −0.81 (−1.42, −0.19) −0.59 (−1.30, 0.11) CA/AA 27 0.69 (−0.58, 1.95) 1.36 (0.04, 2.67) Interaction p-value 0.026 0.009 rs2070673 TT 76 −0.69 (−1.31, −0.06) −0.55 (−1.32, 0.22) TA/AA 37 −0.01 (−1.28, 1.27) 0.53 (−0.64, 1.69) Interaction p-value 0.314 0.116 rs2070676 CC 86 −0.72 (−1.32, −0.13) −0.60 (−1.32, 0.13) CG/GG 27 0.86 (−0.67, 2.39) 1.20 (−0.08, 2.48) Interaction p-value 0.043 0.014 rs2031920 CC 104 −0.41 (−0.97, 0.16) −0.10 (−0.74, 0.55) CT/TT 9 −2.31 (−5.50, 0.88) −1.01 (−4.45, 2.43) Interaction p-value 0.181 0.503 rs2515641 CC 87 −0.80 (−1.40, −0.20) −0.70 (−1.42, 0.03) CT/TT 26 1.32 (−0.12, 2.77) 1.59 (0.41, 2.77) Interaction p-value 0.008 0.002⁎ GSTZ1 (glutathione S-transferase zeta 1) rs1046428 CC 67 −0.34 (−1.05, 0.36) −0.002 (−0.79, 0.79) CT/TT 46 −0.87 (−1.94, 0.20) −0.49 (−1.70, 0.72) Interaction p-value 0.443 0.529 rs7972 GG 95 −0.46 (−1.04, 0.11) −0.19 (−0.86, 0.49) GA/AA 18 −0.52 (−2.63, 1.59) 0.11 (−1.83, 2.05) Interaction p-value 0.952 0.729 rs7975 GG 48 −0.28 (−1.14, 0.58) −0.004 (−1.00, 0.99) GA/AA 65 −0.55 (−1.31, 0.20) −0.18 (−1.02, 0.66) Interaction p-value 0.643 0.795 ⁎ Significant p-value after Bonferroni correction, p-value at 0.004. References Espín-Pérez, A., Font-Ribera, L., van Veldhoven, K., Krauskopf, J., Portengen, L., Chadeau-Hyam, M., et al., 2017. Blood transcriptional and microRNA responses to short-term exposure to disinfection by-products in a swimming pool. Environ Int Abramsson-Zetterberg, L., Zetterberg, G., Bergqvist, M., Grawé, J., 2000. Human cyto- 2018 110, 42–50. genetic biomonitoring using flow-cytometric analysis of micronuclei in transferrin- Fenech, M., 2007. Cytokinesis-block micronucleus cytome assay. Nat. Protoc. 2, positive immature peripheral blood reticulocytes. Environ. Mol. Mutagen. 36, 22–31. 1084–1104. Ainsworth, B.E., Haskell, W.L., Herrmann, S.D., Meckes, N., Bassett Jr., D.R., Tudor- Fernández-Luna, Á., Gallardo, L., Plaza-Carmona, M., García-Unanue, J., Sánchez- Locke, C., et al., 2011. Compendium of physical activities: a second update of codes Sánchez, J., Felipe, J.L., et al., 2012. Respiratory function and changes in lung epi- and MET values. Med. Sci. Sports Exerc. 43, 1575–1581. thelium biomarkers after a short-training intervention in chlorinated vs. ozone indoor ASPB, Agència de Salut Pública de Barcelona, 2012. Quality of Drinking Water in pools. PLoS One 8, e68447. Barcelona. Barcelona. Available at. http://www.aspb.cat/quefem/docs/Aigua_ Font-Ribera, L., Kogevinas, M., Zock, J.P., Gómez, F.P., Barreiro, E., Nieuwenhuijsen, consum_huma.pdf. M.J., et al., 2010. Short-term changes in respiratory biomarkers after swimming in a Bonassi, S., El-Zein, R., Bolognesi, C., Fenech, M., 2011. Micronuclei frequency in per- chlorinated pool. Environ. Health Perspect. 118, 1538–1544. ipheral blood lymphocytes and cancer risk: evidence from human studies. Font-Ribera, L., Villanueva, C.M., Schmalz, C., Zwiener, C., Marco, E., Grimalt, J.O., et al., Mutagenesis 26, 93–100. 2016. Environmental and personal determinants of the uptake of disinfection by- Borg, G., 1970. Perceived exertion as an indicator of somatic stress. Scand. J. Rehabil. products during swimming. Environ. Res. 149, 206–215. Med. 2, 92–98. Gängler, S., Charisiadis, P., Seth, R., Chatterjee, S., Makris, K.C., 2018. Time of the day Broeckaert, F., Clippe, A., Knoops, B., Hermans, C., Bernard, A., 2000. Clara cell secretory dictates the variability of biomarkers of exposure to disinfection byproducts. Environ. protein (CC16): features as a peripheral lung biomarker. Ann. N. Y. Acad. Sci. 923, Int. 112, 33–40. 68–77. Gulati, M., Shaw, L.J., Thisted, R.A., Black, H.R., Merz, C.N.B., Arnsdorf, M.F., 2010. Cantor, K.P., Villanueva, C.M., Silverman, D.T., Figueroa, J.D., Real, F.X., Garcia-Closas, Heart rate response to exercise stress testing in asymptomatic women: the St. James M., et al., 2010. Polymorphisms in GSTT1, GSTZ1, and CYP2E1, disinfection by- women take heart project. Circulation 122, 130–137. products, and risk of bladder cancer in Spain. Environ. Health Perspect. 118, Hery, M., Hecht, G., Gerber, J.M., Gendre, J., Hubert, G., Rebuffaud, J., 1995. Exposure to 1545–1550. chloramines in the atmosphere of indoor swimming pools. Ann Occup Hyg 39, Carbonnelle, S., Francaux, M., Doyle, I., Dumont, X., de Burbure, C., Morel, G., et al., 427–439. 2002. Changes in serum pneumoproteins caused by short-term exposures to nitrogen Holland, N., Davé, V., Venkat, S., Wong, H., Donde, A., Balmes, J.R., et al., 2015. Ozone trichloride in indoor chlorinated swimming pools. Biomarkers 7, 464–478. inhalation leads to a dose-dependent increase of cytogenetic damage in human Carbonnelle, S., Bernard, A., Doyle, I.R., Grutters, J., Francaux, M., 2008. Fractional lymphocytes. Environ. Mol. Mutagen. 56, 378–387. exhaled NO and serum pneumoproteins after swimming in a chlorinated pool. Med. International HapMap Consortium, 2003. The International HapMap project. Nature 426, Sci. Sports Exerc. 40, 1472–1476. 789–796. Cardador, M.J., Gallego, M., 2011. Haloacetic acids in swimming pools: swimmer and Kissling, G.E., Dertinger, S.D., Hayashi, M., MacGregor, J.T., 2007. Sensitivity of the er- worker exposure. Environ Sci Technol 45, 5783–5790. ythrocyte micronucleus assay: dependence on number of cells scored and inter-an- Chowdhury, S., Alhooshani, K., Karanfil, T., 2014. Disinfection byproducts in swimming imal variability. Mutat. Res. 634, 235–240. pool: occurrences, implications and future needs. Water Res. 53, 68–109. Kogevinas, M., Villanueva, C.M., Font-Ribera, L., Liviac, D., Bustamante, M., Espinoza, F., Costa, C., Silva, S., Neves, J., Coelho, P., Costa, S., Laffon, B., et al., 2011. Micronucleus et al., 2010. Genotoxic effects in swimmers exposed to disinfection by-products in frequencies in lymphocytes and reticulocytes in a pesticide-exposed population in indoor swimming pools. Environ. Health Perspect. 118, 1531–1537. Portugal. J Toxicol Environ Health A 74, 960–970. Llana-Belloch, S., Priego Quesada, J.I., Pérez-Soriano, P., Lucas-Cuevas, Á.G., Salvador- Erdinger, L., Kühn, K.P., Kirsch, F., Feldhues, R., Fröbel, T., Nohynek, B., et al., 2004. Pascual, A., Olaso-González, G., et al., 2015. Disinfection by-products effect on Pathways of trihalomethane uptake in swimming pools. Int. J. Hyg. Environ. Health swimmers oxidative stress and respiratory damage. Eur. J. Sport Sci. 12, 1–9. 207, 571–575. Marco, E., Lourencetti, C., Grimalt, J.O., Gari, M., Fernández, P., Font-Ribera, L., et al., 10 L. Font-Ribera, et al. Environment International 131 (2019) 104988 2015. Influence of physical activity in the intake of trihalomethanes in indoor Perspect. 122, 100–106. swimming pools. Environ. Res. 140, 292–299. Stockfelt, L., Sallsten, G., Olin, A.C., Almerud, P., Samuelsson, L., Johannesson, S., et al., Nordberg, G.F., Lundstrom, N.G., Forsberg, B., Hagenbjork-Gustafsson, A., Lagerkvist, 2012. Effects on airways of short-term exposure to two kinds of wood smoke in a B.J., Nilsson, J., et al., 2012. Lung function in volunteers before and after exposure to chamber study of healthy humans. Inhal. Toxicol. 24, 47–59. trichloramine in indoor pool environments and asthma in a cohort of pool workers. Stopper, H., Hempel, K., Reiners, C., Vershenya, S., Lorenz, R., Vukicevic, V., et al., 2005. BMJ Open 2, e000973. Pilot study for comparison of reticulocyte-micronulei with lymphocyte-micronuclei Plewa MJ, Wagner ED, Muellner MG, Hsu KM, Richardson SD. 2008. Comparative in human biomonitoring. Toxicol. Lett. 156, 351–360. mammalian cell toxicity of N-DBPs and C-DBPs, in: T. Karanfil, S.W. Krasner, P. Thomas, P., Holland, N., Bolognesi, C., Kirsch-Volders, M., Bonassi, S., Zeiger, E., et al., Westerhoff, Y. Xie (Eds.), Occurrence, Formation, Health Effects and Control of 2009. Buccal micronucleus cytome assay. Nat. Protoc. 4, 825–837. Disinfection By-products in Drinking Water. American Chemical Society, van Veldhoven, K., Keski-Rahkonen, P., Barupal, D.K., Villanueva, C.M., Font-Ribera, L., Washington, DC, Symposium Series No. 995, 36. Scalbert, A., et al., 2018. Effects of exposure to water disinfection by-products in a Provost, E.B., Chaumont, A., Kicinski, M., Cox, B., Fierens, F., Bernard, A., Nawrot, T.S., swimming pool: a metabolome-wide association study. Environ. Int. 111, 60–70. 2014. Serum levels of club cell secretory protein (Clara) and short- and long-term Villanueva, C.M., Font-Ribera, L., 2012. Health impact of disinfection by-products in exposure to particulate air pollution in adolescents. Environ. Int. 68, 66–70. swimming pools. Ann. Ist. Super. Sanita 48, 387–396. Richardson, S.D., Plewa, M.J., Wagner, E.D., Schoeny, R., DeMarini, D.M., 2007. Villanueva, C.M., Cantor, K.P., Cordier, S., Jaakkola, J.J., King, W.D., Lynch, C.F., et al., Occurrence, genotoxicity, and carcinogenicity of regulated and emerging disinfection 2004. Disinfection byproducts and bladder cancer: a pooled analysis. Epidemiology by-products in drinking water: a review and roadmap for research. Mutat. Res. 636, 15, 357–367. 178–242. Villanueva, C.M., Cantor, K.P., Grimalt, J.O., Malats, N., Silverman, D., Tardon, A., et al., Richardson, S.D., DeMarini, D.M., Kogevinas, M., Fernandez, P., Marco, E., Lourencetti, 2007. Bladder cancer and exposure to water disinfection by-products through in- C., et al., 2010. What's in the pool? A comprehensive identification of disinfection by- gestion, bathing, showering, and swimming in pools. Am. J. Epidemiol. 165, products and assessment of mutagenicity of chlorinated and brominated swimming 148–156. pool water. Environ. Health Perspect. 118, 1523–1530. Vlaanderen, J., van Veldhoven, K., Font-Ribera, L., Villanueva, C.M., Chadeau-Hyam, M., Stayner, L.T., Pedersen, M., Patelarou, E., Decordier, I., Vande Loock, K., Chatzi, L., et al., Portengen, L., et al., 2017. Acute changes in serum immune markers due to swim- 2014. Exposure to brominated trihalomethanes in water during pregnancy and mi- ming in a chlorinated pool. Environ. Int. 105, 1–11. cronuclei frequency in maternal and cord blood lymphocytes. Environ. Health 11